Re: [Teas] Review of draft-ietf-teas-5g-ns-ip-mpls

["Wubo (lana)" <[e-mail]>]

Hi Krzysztof,

Thanks for the reply. Please see inline.

Regards,
Bo Wu

From: Krzysztof Szarkowicz <kszarkowicz=[e-mail]>
Sent: Thursday, December 7, 2023 1:47 AM
To: Wubo (lana) <[e-mail]>
Cc: TEAS WG <[e-mail]>; Richard Roberts <[e-mail]>; Mr. Mohamed Boucadair <[e-mail]>; Julian Lucek <[e-mail]>; Luis Miguel Contreras Murillo <[e-mail]>
Subject: Re: [Teas] Review of draft-ietf-teas-5g-ns-ip-mpls

Hi Bo Wu,

Please see inline.

Thanks,
Krzysztof



On 2023 -Dec-03, at 04:39, Wubo (lana) <lana.wubo=[e-mail]<mailto:lana.wubo=[e-mail]>> wrote:

Hi Authors, WG,

I have read draft-ietf-teas-5g-ns-ip-mpls and have several general comments:

1. On the mapping between IETF Network Slice Service SDP and 5G CE, this document and draft-5g-network-slice-application (5g-application) both provide some content. I think it would be more clear to list all 5G CE options in one document and avoid the inconsistency between the two documents. For example, Section 4 Overview of the Realization Model describes three options: VLAN, IP, and MPLS. However draft-5g-application has more options than these. This document says that 5G CEs have BGP protocol mapping, while Section 5.3 of draft-5g-application does not .

[Krzysztof] As per agreement between 5G application and 5G realization draft, the hand-off options between different domains is part of realization. So, are you suggesting that hand-off options should be moved from 5G-application to 5G-realization draft, and the text combined in the 5G-realization draft? Regarding “This document says that 5G CEs have BGP protocol mapping, ..” could you refer to specific section?

[Bo Wu] Yes, It is suggested that the hand-off options can be described in one document to keep consistency, as my understanding is that these options are related to SDP AC mapping and need to be clarified in 5G application. For“This document says that 5G CEs have BGP protocol mapping, ..”,Section 3.3.3 (https://datatracker.ietf.org/doc/html/draft-ietf-teas-5g-ns-ip-mpls-02#name-service-aware-ce) and 4.3.2.indicate that there is BGP control plane mapping. But https://datatracker.ietf.org/doc/html/draft-ietf-teas-5g-network-slice-application-02#section-5.3 says no control plane interaction.


2. Section 6 Transport Planes Mapping Models introduces "tunnel group" or "transport planes'. Can you clarify how to manage the tunnel groups?
Does tunnel group management also requires topology and resource allocation management? As I am wondering the difference between tunnel group management and NRP management. Per Section 4.2. Control Plane Network Resource Partition Mode of draft-ietf-teas-ns-ip-mpls, NRP has a mode of control plane without data plane encapsulation mechanism.


[Krzysztof] “Tunnel groups” or “transport planes” is the concept deployed in production networks by many operators since years, where multiple TE tunnels are created between PEs, using different path optimization method. For example, some tunnels are optimization for latency, some other tunnels are just following default IGP path, etc. The way how these tunnels are managed is out-of-scope for this document, but we provide some examples, like Flex-Algo or RSVP/SR traffic engineering tunnels with or without PCE, and with or without bandwidth reservations. Further, Section 7, especially section 7.2, discusses different capacity planning and bandwidth models for these tunnels.

[Bo Wu] To me, tunnel group seems like also an implementation of NRP, representing part of resources of the underlay network. As per the definition, “NRP is defined as a collection of resources identified in the underlay network”.

3. Section 3 5G Network Slicing Integration in Transport Networks defines various distributed CE-PE models which are not used in subsequent sections.

[Krzysztof] Section 3.3.1 has following text at the end: “In subsequent sections of this document, the terms CE and PE are used for both single and distributed devices."


Are these different models related to the implementation of QoS mapping, connectivity, or capacity planning? Could you add some text to reflect that?

[Krzysztof] Not sure, if I follow exactly what is the request here. QoS mapping, transport plane mapping, or capacity planning is orthogonal to the fact, if the PE is distributed or not.

[Bo Wu] I agree with you on this. It seems that the distributed options of CEs and PEs are irrelevant to the slice realization in chapters 4 and 5. Since it is not closely related to slice realization, Section 3 may go into appendix as some examples or merge with draft-5g-application.

Thanks,
Bo Wu
_______________________________________________
Teas mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/teas